The acquisition of spatial data from archival photographs and their application to geomorphology
This thesis discusses the development and application of an analytical photogrammetric technique which enables accurate spatial data, of known quality, to be derived from archival photographs. Such a facility represents an important advancement, particularly for geomorphologists, because the effects of geomorphological process can be assessed quantitatively and directly by comparing spatial data derived from photographs at different epochs. Sources of archival photographs of England are identified and the type, quantity, range and age of each major collection is discussed. Existing methods of deriving spatial data from photographs are reviewed and illustrated by previous research, with particular emphasis upon the limitations associated with each method. The technique that was developed is based upon a self calibrating bundle adjustment and both the functional and stochastic models suitable for successful restitution of archival photographs were established. Five computer programs were developed and the algorithms associated with each are given. These programs are run sequentially and assist in rapid restitution of archival photography and to derive measures of data quality. The technique is applied successfully to a forty year old sequence of archival photographs, obtained from a variety of sources, of the Black Ven landslide, Dorset, England. Spatial data was derived from five photographic epochs, at approximately 10 year intervals, using an analytical plotter. A secondary aim of the research was to extend existing techniques and devise new methods of processing these spatial data, for geomorphological purposes. Several techniques were found to be especially valuable including: the production of morpho-genetic maps; DTI's of difference; evolutionary models; animated sequences and distributions of slope angle. The latter has shown that the evolutionary model of 'dynamic equilibrium is valid for the Black Ven landslides. All aspects of data quality are examined, particularly the functional model used in the self-calibrating bundle adjustment. This least squares estimating procedure is found to be perfectly adequate for the successful restitution of archival photographs.